Predictive Analysis of PM2.5 Concentration in Taiyuan City Based on Deep Learning

With the continuous development of China's economy, the air pollution problem has become increasingly serious, and PM2.5 concentration has become an important indicator of the degree of air pollution. Accurate prediction of PM2.5 concentration is of great significance for public health and ecological environment protection. In this paper, based on deep learning technology, LSTM and GRU models were selected to combine the meteorological data and air quality data of Taiyuan City in 2023 to conduct a prediction study of PM2.5 concentration. Through data preprocessing, feature selection and model construction, the prediction results of the two models were compared and analysed using assessment indicators such as MAE, RMSE, MBE and R². The experimental results show that the GRU model is slightly better than the LSTM model in terms of prediction accuracy and response to short-term variations, but the prediction results of the two models still have some errors, and more comprehensive feature variables should be considered in future related studies.